Core Minutes 12/9/2008

• GR status:

  • Tags, system tests (Richard) System tests have been done for LPM effect; C & A is still scratching its collective head (no problems seen yet). Next up: interactive alphas. (Heather) believes Franz still needs to make some updates to a couple packages.

    (Heather) GR v15r52p1 has the upgrade to LDF! And has been run through systests.

  • Overlay (Richard) The code has been committed and is in a tagged GR, ready for System Tests. (Tracy) Leon pointed out we're not running with the correct values for splits. Suggest we first do a run with just that change; then add the overlays. (Leon) We have 434K overlay events to play with. (Richard) asks that needs for new IRFs be discussed at the next C & A meeting; (Philippe) will put it on the agenda.

    (Leon) Mentioned that Tracy had found a bug in RootTupleSvc, but did not feel qualified to describe it in detail, so (Tracy) sent this for inclusion in the minutes:

    During the "begin event" calll in RootTupleSvc, there is a loop over input trees/branches with a call to "GetEvent" for each. The argument for this call is the desired event number, unfortunately, the call (inside the loop) increments the counter. So... the first call gets the right event, after that you are lost. In addition merit is filled per event so makes sense to "get" each event, but the pointing info is not filled every event so shouldn't be "gotten" each event. I think. Heather is the expert. [It seems no one is anxious to claim this one.]

  • Reprocessing readiness (Tom) There are two parts. The task for pre-Sept. 2 data is essentially ready to go. The second part is post-Sept 2 data: just redoing the FT1. That has yet to be assembled but should be easy.

• FSSC Report: (Eric W.) [kindly provided the following; coming also to a ScienceTools report near you]

  Once again, not much to report from the FSSC. Dave is working on test scripts and data that we will use for our brutal but thorough testing regime in preparation for our release on 15 February. John is collecting and integrating all of the various test programs from the SLAC-derived Science Tools code tree for use in a quick-check build-time test (i.e. a "hmake test" target). James has been executing a variety of bug fixes. And I've been running down problems getting pyLikelihood to build and install properly under the HEADAS system, which has forced me to take the plunge and learn a lot about how SWIG works when gluing C++ code to Python.

  That's all from the East Coast

• ScienceTools: (Jim) Highlights are:

  • New release of ScienceTools, v9r9, includes the new diffuse response columns. The code can read old as well as new files, but cannot add the columns to old files.
  • People were reporting problems with large files, which turned out to be due to a bug in cfitsio. We have a patch.
  See also the full report.

• Documentation: (Chuck) sends this summary:

  In addition to routine updates, highlights in WB World include completion of the first round of the Data Access Help files. However, I still need to update or generate new help files for the key monitoring tools (pipeline, Data Processing, ASP Data Viewer, etc.). Yesterday, I received Masa's go-ahead to incorporate his updates to some of the Pulsar Tools Tutorials, and will incorporate them in the WB today or tomorrow.

  I'm also working on a major top-down restructuring of the workbook, and have completed extensive changes to the green, top-level Data Access navbar and to the level-2 navbars linked from the Monitoring Tools and Help Files buttons

  Substantial progress has also been made on revising the rednavbar section, but these changes are not yet ready to post to the web, though I hope to complete them before the end of the year.

  In the meantime, I'm working on this month's newsletter, due out later this week.

• Core talk:

  • skimmer (Heather) You may have noticed a lot of skimmer JIRA activity recently with various items being marked "fixed". They're not really fixed, just moved to a different bug tracker. Since the code is no longer a Glast-maintained product, having been moved by David (Chamont) to his own Subversion respository, it makes sense that the bug tracking should move also. David will continue to install new versions of the skimmmer at SLAC as it evolves and we are of course free to patch locally if necessary (and forward the patch to David for inclusion in his next release).

  • ROOT 5.20 (Heather) The distribution would not compile with gcc 3.2.3. It is now patched and a build is in progress.

  • RM web interface (Heather for Karen) is now accessible outside the firewall at http://glast-ground.slac.stanford.edu/rm2/

  • Navid report

    • Tag collector He ran into a bizarre snag on Windows only where a certain piece of code is not getting invoked, even though it behaves properly on Linux and Mac. He thinks he has a work-around; a new version should be available soon, probably today, with much faster querying.
    • RM on RHEL5 Our RHEL5 boxes still don't have MySQL installed, hence RM can't run there.
    • RM on Mac Awaiting installation of lsf on these machines.
    • Override Supersede After a bi-coastal debugging session last week Navid and Joanne believe the supersede functionality (renamed since the term override was already reserved for something else) is behaving as advertised.

  • GR and Externals (Emmanuel) Updates needed to build GR v15r49 with SCons have been committed to CVS.   New or revised SCons-related files in each of the affected packages are now properly tagged with the correct package tag and tag for GR v15r49.  He has some questions on how to proceed with tagging other (not SCons-mandated) files, to be resolved in a post-meeting discussion with Heather.

  • GoGui (Joanne) Main news is GoGui support for supersede. There are some situations in which the GoGui display needs to tell the user a little more — in particular, it should clearly indicate whenever an SCons init operation is called for; or else cancel the currently-running SCons process, which effectively forces a new init — but the basic supersede functionality is there. Users are likely to notice some, possibly irritating, differences between SCons supersede and handling of multiple paths by CMT. The SCons behavior is not necessarily bad or illogical, but may take some getting used to.

    She'll take a short break from GoGui work for the next few days to finish up the work Bryson has done to sconsify CHS. Remaining packages are MOOT-related. One of them, LATC_vrfy was going to be particularly problematic since it needs access to non-public fsw files, but there is now a better way to deal with this. Owen Saxton has replaced the LATC parser with a new fsw facility, QCFG, which does everything the old facility + LATC_vrfy used to do; however using it will require non-trivial changes in the MOOT package (included in CHS and used by Online to generate new configurations; not included in GlastRelease).

    Then it will be back to GoGui on Windows support.

• Announcements: (Richard) Tracy got an offer he couldn't refuse to go sailing in Florida. He leaves Thursday and will return Monday.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:46:20 -0700